Form ApprovedREPORT DOCUMENTATION PAGE OMB No. 0704-0188

Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,gathering and maintaiiing the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of thiscollection of inform':ion, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 JeffersonDavis Hiphway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.

1. AGENCY USE ONLY (Leave 2. REPORT DATE 3. REPORT TYPE AND DATES COVEREDBlank) March 31, 2007 Final Technical Report, June 2001-November 30, 2006.

4. TITLE AND SUBTITLE 5. FUNDING NUMBERSBasic Research Leading to Compact, Portable Pulsed Power F496200110354FY'01 MURI Consortium

6. AUTHORSEdl Schamiloglu, Karl Schoenbach and Robert Vidmari

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORTUniversity of New Mexico NUMBER

Department of Electrical and Computer EngineeringMSC01 1100

Albuquerque, NM 87131-00-1

9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING / MONITORING AGENCYAFOSR/NE875 North Randolph St., Suite 325Arlington, VA 22203-1977 AFRL-SR -AR-TR"-7-0119Dr. Robert J. Barker, PM

11. SUPPLEMENTARY NOTES

12a. DISTRIBUTION /AVAILABILITY STATEMENT 12b. DISTRIBUTION CODEUnlimited AFOSPJNE

13. ABSTRACT (Maximum 200 words)Pulsed power is a technology that is suited to drive electrical loads requiring very large power pulses in short bursts (high-peak power). Certain

applications require technology that can be deployed in small spaces under stressful environments, e.g., on a ship, vehicle, or aircraft. In 2001, the U.S.Department of Defense (DoD) launched a long-range (five-year) Multidisciplinary University Research Initiative (MURI) to study fundamental issues forcompact pulsed power. This research program endeavored to: 1) introduce new materials for use in pulsed power systems; 2) examine alternativetopologies for compact pulse generation; 3) study pulsed power switches, including pseudospark switches; and 4) investigate the basic physics relatedto the generation of pulsed power, such as the behavior of liquid dielectrics under intense electric field conditions. This final technical report reviews theadvances put forth by the researchers in this consortium and will assess the potential impact for future development of compact pulsed power systems.The progress made over the final 18 months of the program is emphasized. The key achievements of this research program are: i) liquid breakdownhas been extensively diagnosed and new models to explain the observed behavior have been proposed; ii) dielectric breakdown under pulsed powerconditions was observed to occur at a lower threshold (about 75%) compared with quasi-DC breakdown for a wide range of materials ranging fromcommon Mylar to complicated high dielectric nanopowder-impregnated expoy resin - a new model to explain this behavior for nanocrystalline TiO2 hasbeen proposed; fundamental limitations in achieving compact, folded planar Blumlein transmission lines have been identified; the investigation of a100-to-1 impedance ratio exponential line demonstrated an approach to maintaining coherent phase fronts at the wide low-impedance end of anexponential line - fundamental limitations to this approach, as well as the use of turbulent capillary flow for mitigating thermal loading were identified.

14. SUBJECT TERMS 15. NUMBER OF PAGESPulsed power, compact pulsed power, Blumlein, electromagnetic modeling, high pressure switches, liquid 34breakdown, diectric breakdown, thermal management.

16. PRICE CODE

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACTOF REPORT OF THIS PAGE OF ABSTRACT NoneUnclassified Unclassified Unclassified

NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)Prescribed by ANSI Std. Z39-1298-102

The Unrkhetr.y of Ncw \Ixico I

The University of New MexicoPulsed Power, Beams and Microwaves Laboratory,Department of Electrical & Computer Engineering

FY'01 MURI Consortium:Basic Research Leading to Compact, Portable Pulsed Power

Final Technical Report1 June 2001 - 30 November 2006

31 March 2007

Submitted by:

Edl Schamiloglu - PI Karl H. Schoenbach Robert VidmarGardner-Zemnke Professor, Eminent Scholar, Batten Research Professor,Department of Electrical Endowed Chair in Physics Department,and Computer Bioelectric Engineering, University Of Nevada,Engineering Frank Reidy Research RenoUniversity of New Mexico Center for Bioelectrics Physics Room 206, 5625Albuquerque, NM 87131 Old Dominion University Fox Ave, Reno, NV 89506Tel. (505) 277-4423 830 Southampton Avenue,Fax: (505) 277-1439 Suite 5100edl@ece .unm. edu Norfolk, VA 23510

20070417194

TABLE OF CONTENTS

EXECUTIVE SUM M ARY ................................................................................................ 4UNM - LIST OF ACHIEVEMENTS ................................................................................ 5

Research in Support of the Development of a Compact Blumlein Pulser Test Bed ........ 5B ackground: .......................................................................................................... 5A ccom plishm ents: ..................................................................................................... . 5Liquid Breakdown Studies in Support of Task 1 ................................................... 14

High Dielectric Constant Ceramics for Use in Blumlein Transmission Lines .............. 14UNM PERSONNEL ........................................................................................................... 16UNM PUBLICATIONS .......................................................................... 17UNM RECOGNITION ...................................................................................................... 19UNM - NEW DISCOVERIES, INVENTIONS, PATENTS .......................................... 19ODU INTRODUCTION .................................................................................................... 21ODU - LIST OF ACHIEVMENTS ................................................................................ 21

High repetition-rate, compact, nanosecond pulse generator .......................................... 22Electrical Triggering of W ater Switches ....................................................................... 23

ODU PUBLICATIONS .......................................................................... 25UNR ABSTRACTIONS... ................................................... ............................................. 26UNR INTRODUCTION ....................................................................................................... 28UNR - LIST OF ACHIEVEMENTS ................................................................................... 29

A. Exponential Transmission Line M odeling ............................................................... 29B . RF Pow er Term ination .............................................................................................. 29

UNR - SUMMARY OF RESULTS ................................................................................ 31A. Limitations in Implementation of Exponential Line Theory .................................... 31B. Results from an Exponential Line with a 100-to-I Impedance Ratio ....................... 31C. Utilization of RF Chip Termination Resistors .......................................................... 32

UNR PERSONNEL............................................................................... 33UNR PUBLICATIONS ................................................................................................... 33UNR REFERENCES ........................................................................................................ 34

3

EXECUTIVE SUMMARY

Pulsed power is a technology that is well-suited to driving electrical loads requiring verylarge power pulses in short bursts (high-peak power). Certain applications require technologythat can be deployed in small spaces under stressful environments, e.g., on a ship, vehicle, oraircraft. In 2001, the U.S. Department of Defense (DoD) launched a five-yearMultidisciplinary University Research Initiative (MURI) to study fundamental issues forcompact pulsed power. This research program endeavored to: 1) introduce new materials foruse in pulsed power systems; 2) examine alternative topologies for compact pulse generation;3) study pulsed power switches, including pseudospark switches; and 4) investigate the basicphysics related to the generation of pulsed power, such as the behavior of liquid dielectricsunder intense electric field conditions. Furthermore, the integration of all of these buildingblocks is impacted by system architecture (how things are put together). The mainapplication for which this MURI program received its motivation from is intense beam-driven sources of high power microwaves (HPM). (A sister consortium led by USC alsocontributed to this overall program.)

After the three-year mid-point review the DoD advisory panel recommended that theCompact Pulsed Power MURI teams focus their efforts on liquid breakdown as the maintechnical challenge. Other research topics continued in parallel, albeit with diminishedemphasis.

This final technical report summarizes the critical advances put forth by the researchers inthis consortium. The emphasis of this report is on the last 18 months of the program.Technical summaries of earlier efforts, as well as a listing of publications, can be found inthe yearly annual reports that were submitted. For ease of exposition, the report is organizedby the participating universities. Old Dominion University was the lead on the liquidbreakdown studies.

The key achievements of this research program are: i) liquid breakdown (for water andpropylene carbonate) has been extensively diagnosed and new models to explain theobserved behavior have been proposed; ii) dielectric breakdown under pulsed powerconditions was observed to occur at a lower threshold (about 75%) compared with quasi-DCbreakdown for a wide range of materials ranging from common Mylar to complicated highdielectric nanopowder-impregnated expoy resin - a new model to explain this behavior fornanocrystalline TiO2 has been proposed; fundamental limitations in achieving compact,folded planar Blumlein transmission lines have been identified; the investigation of a 100-to-1 impedance ratio exponential line demonstrated the feasibility of an approach to maintainingcoherent phase fronts at the wide low-impedance end of an exponential line - fundamentallimitations to this approach, as well as the use of turbulent capillary flow for mitigatingthermal loading were identified.

4

UNM - LIST OF ACHIEVEMENTS

Research in Support of the Development of a Compact Blumlein PulserTest Bed

Background:

The research described in this subsection was to support the development of a 600 kV, 50 ns,5 Q compact pulse-shaping drive that would be significantly lighter than coaxial watertransmission line systems. Whereas researchers in the past have studied stacked-Blumleinarrangements using coaxial cable,1 it is anticipated that a higher energy density (andtherefore more compact) system can be achieved using a folded stripline geometry. Studiesby Alexander et al. at Sandia National Laboratories2 demonstrated that propylene carbonate-impregnated Kapton TM sheets are attractive for constructing high average field striplines thatcan be folded into a compact geometry. Such a compact source can be used to drive lowimpedance HPM sources (such as a Vircator, MILO, or relativistic magnetron) on a portableplatform.

Accomplishments:

The initial Blumlein transmission line configuration that is being studied is the systemdescribed earlier.2 The pulser that was used to charge the pulse-shaping section is a 6-stagebipolar charged Marx system. The capacitors in the Marx are 0.2 laF, 100 kV cans and theswitches are MaxwellTM midplane-triggered pancake spark gaps. The pulse shaping sectioncomprises the Blumlein transmission line and its switch (triggered rail gap).

The Blumlein transmission line section utilized 10" wide copper strips with 2" Kapton TM

margins in a 48" line that was folded into a 15" long unit. The dimensions were selected toachieve the design line impedance Z utilizing

377 *d Z • *, /•-- •'(1)

where d is the thickness of the dielectric medium, w is the width of the line (assuming d/w <0.1), and cr is the dielectric constant. The parameters utilized yield an impedance of 5 0).

Electromagnetic Modeling in Support of This Task:

The approach at UNM was to first model the actual folded Blumlein configurationelectromagnetically. In Fig. 1 we present the results of an HFSSTM (Agilent/Ansoft)

SSee, for example, I.C. Somerville, S.J. MacGregor, and 0. Farish, "An Efficient Stacked-Blulmein HV Pulse

Generator, Meas. Sci. Techno. 1, 865 (1990) and references therein.2 J.A. Alexander, S. Shope, R. Pate, L. Rinehart, J. Jojola, M. Ruebush, W. Crowe, J. Lundstrom, T. Smith, D.

Zagar, and K. Prestwich, "Plastic Laminate Pulsed Power Development," Proceedings, Society of AutomotiveEngineers, Paper 0OPSC-113 (San Diego, CA, October, 2000).

5

calculation to explore the electric field distribution between the center and outer conductorsof the Blumlein configuration. In this configuration a voltage waveform is launched betweenthe top and center conductors. A portion of the waveform is transmitted and propagatesbetween the center conductor and the bottom conductor. A portion of the waveform reflectsback upstream in the top section. Finally, after a double transit time, the full line chargevoltage appears at the load (shown to be at the end of the line where the coordinate axes aredefined in the sketch).

Figure 1. Calculation of voltage distribution at the load (defined by the position of thecoordinate axes in the sketch). The red indicates full charge voltage at the load.

The importance of such calculations is to ultimately include the effects of resistivity gradingat the outer ends of the line (where the dielectric and liquid interact) and to propose tailoreddielectrics.

The research on this topic finally focused on using CST Microwave Studio as the simulationtool of choice. Two issues were studied: i) the effects that multi-switch excitation has on theresponse of a parallel-plate Blumlein line, and ii) the effects of using a multi-layer dielectricin a parallel-plate Blumlein pulse-forming line.

Effects of Multi-Switch Excitation

It is known that the use of multiple switches achieves faster rise time by reducing the overallinductance. 3 Total inductance is inversely proportional to the number of switches in thesystem. In this work we examine where the practical limit to this is, i.e. what the number ofswitches is when no further change is observed if the number of switches is increased.However, in this work, we do not model the switch using its inductance, but rather by usingthe voltage transition produced by switch triggering. All the results here are produced usingCST Microwave StudioTM, a 3D-electromagnetic solver.

To better observe the effects of multi-switch excitation, we applied to a few model lines of

3 S.T. Pai and Q. Zhang, Introduction to High Power Pulse Technology, World Scientific, 1995.

6

different widths. We compared the line response to a growing number of switches in eachmodel in terms of the pulse width on the load, and maximum electric field. Following that,the pulse robustness against the imperfect triggering of the multi-switch system was shown(i.e., a couple of delayed-triggering models). Ultimately, we test the use of smaller spacingbetween the switches near the edge of the line to see if that will further improve theresponsiveness of the line.

Load Pulse Shapes

For this examination, we used four line widths: a) w = 25.4 cm, b) w = 19.05 cm, c) w = 12.7cm, and d) w = 6.35 cm. It was estimated that case (a) can be tested for up to 7 switches,while cases (b) through (d) will use up to 5 switches. To have a symmetrical distribution ofswitches, odd numbers of them are used, as well as equal spacing between them, with oneswitch always placed in the center of the line width (later shown in Fig. 4a). Also, it wasinitially assumed that all the switches trigger at the same time.

Figure 2 shows a comparison of the pulses created by multiple switches for the above cases(a) and (d). The behavior of cases (b) and (c) is qualitatively and quantitatively in betweenthese two cases.

The following observations can be drawn from these figures: use of a multi-switch excitationshortens the pulse, in general; the greater the number of switches, the shorter the pulse (thiseffect is stronger in a wider line); but, as the number of switches grows, the change in thepulse length is smaller and the pulse exhibits a growing ripple. At that point, the pulse lengthdoes not practically change with the increase in the number of switches, but the ripple does.The ripple is greater if the line is narrower. In addition, the pulse will be slightly shorter (i.e.the edges steeper) for a narrower line.

1.' ~ ~ ~ ~ n _____________

.i . ................

(a)

7

0 ,5 . . . .. . . .. ... . . ... . . .. . .. . . . . .. . . . . .. .. .. .. .. .. ... .. ... .. .. ... ....... .. ... .... ...... .... ..... .... .... • 'h

.1';........ .... ............. ... ...... . . . .

.. .... .. . .

(b)

Figure 2. Blumlein response to a different number of switches for: a) w = 25.4 cm; b) w = 6.35 cm.

Our results qualitatively agree with the observations made about the effects of multipleswitches in an experimental work.4 Both results show somewhat faster saturation thantheoretical analysis would suggest. The theory states that the total inductance of a multi-switch excitation is approximately equal to L = Li/n, where Li is the inductance of a singleswitch, and n is the number of switches, assuming equal switches are utilized. Thus, theinductance should be steadily dropping as the number of switches increases (most noticeablyfor up to 8 switches), which can be verified by plotting the relative change of the totalinductance with respect to a single-switch system. The results in Fig. 2 (and the cases (b) and(c) that are not shown) match that expectation, but saturate quite faster - almost after having3 switches in the system. Therefore, these results suggest that, if a multi-switch system isused, there seems to be no advantage of using more than 3 switches in the system (usingcomparable line widths to these used in cases (b) to (d)).

Effects of Triggering Delay Between Switches

In the above simulations, we modeled all the switches to trigger at the same instant in time.That is an ideal situation to deal with, but a realistic system can exhibit a delay involvedbetween the switches during the triggering process. Here, two cases are studied. In the firstcase, we assume that the switches trigger consecutively from the outermost switch of one endof the line towards the opposite outermost switch, and we refer to this case as a single-sideddelay. For this case, we assign a constant delay of 0.3 ns. In the second case, we assume thatthe switch in the center triggers first, followed by the adjacent switches on each side towardsthe outermost switches. This case is referred to as double-sided delay and the delay timeassigned to it is 0.3 ns as well.

J j. Lehr, "High Power Water Switches for ZR at Sandia National Laboratories," [seminar talk], IEEEAlbuquerque Chapter, January 20, 2005.

8

Response to Exc. wth Delay btw. Switches: w = 25.4 cm, er = 3.5, tau = 10 ns

0.5-

, i if

0 . .. . . . ... ............. .

-0.5 .. -

tiff I V ..... ..C) --------- I ----: :dell'!=0,3 ns, 2-•ided

15 ', d____,,__,_ r_____ -_______

0 1 0 2 0 30

bme, ns

Figure 3. Output pulse after different triggering-delay patterns in the multi-switch excitation.

The results are shown in Fig. 3, compared to a no-delay case. For 0.3 ns-delay, the lineresponse is robust against the triggering-synchronism imperfections of the multi-switchexcitation system, especially for a 2-sided pattern. However, if a delay of 1.5 ns occurred, theresponse would still hold up for the 2-sided delay pattern, but it would be entirely corruptedfor the 1-sided delay case. Overall, these cases give us a pattern and a range of switch-delayvalues for which the line can still produce a regular pulse.

Unequal Spacing Between Switches

Since it was noted that the effects of a multi-switch system weaken with the growing numberof switches, one can suspect that the switches in the central part have more influence in themulti-switch system. To improve that situation, one idea may be to use a higher density ofthe switches near the edges of the line.4 Therefore, we modified our original cases (a), and(d), each comprising their maximum number of switches, by adding switches on each side ofthe excitation system. Four of these configurations were tried, each with different spacingand switch position. If the original separation of the switches is denoted with s, these addedswitches have separations of s/2, s/3, and s/4.

New test configurations in comparison with the "original" one are sketched in Fig. 4 for a 5-switch system. They are analogously set in the 7-switch system. These four testconfigurations should be sufficient if there is any advantage of using an increased switchdensity at the outer part of the line.

9

(a)

(b)

(c)

(d)

(e)

Figure 4. Configurations of a multi-switch system: a) equidistant; b) I switch added outwardly (oneach side) at s/2 distance (called 5+2); c) 2 switches added at s/3 distance each (called 5+4); d) 3switches added at s/3: 2 switches outwardly, and I switch inwardly (called 5+6 in); e) 3 switchesadded outwardly at s/4 each (called 5+6 out).

In addition, the configuration in Fig. 4d makes a more gradual transition from the originalequidistant configuration by inserting one switch between the original switches 1 and 2.

10

............... ... ...

iYif

(a).. /

(b)

Figure 5. Blumlein line response to adding more switches near the edge of the line for line width: a)25.4 cm; b) 6.35 cm.

The result for the line cases (a), and (d) are shown in Fig. 5. In the case of the wider line,these additional switches did shorten the pulse a little (previously mentioned general ruleapplied), by affecting the falling edge of the pulse, but there is practically no difference inresponses produced by the different configurations in Fig. 4. In the case of a narrow line inFig. 5(b), these additional switch combinations showed no difference to the original 5-switchsetting.

Effects of using a Multi-Layer Dielectric

The analyses were performed using the CST Microwave Studio software on a scaled downversion of a straight, parallel-plate Blumlein line with the following parameters: the width w= I cm, the length L = 3 cm, the permittivity of the dielectric Er = 30, the thickness of thedielectric slab d = 0.762 cm. The size was chosen to reduce the amount of computation time

11

r equired for a full scale model. The choice of 15 and 5 layers with this separation ofelectrodes provides good paradigms and "rounded" numbers for the thickness of each layer,which enables accurate modeling. Here we report the results from the 15-layer model.

The conductivity function was derived based on the following principle: minimum andmaximum conductivity values were first and, depending on the actual number of layers usedin the design, the remaining values of conductivity were then derived such that equalincrements (steps) between the minimum and maximum value were obtained. The electricfield (E-field, for short) inside the layers was recorded using a probe that was inserted in thevertical center of each layer at the edge of the line, where the maximum fields are reported tooccur [10-12].

15-layer Model

This design comprises 15 equally thick layers of the layer thickness dl = d/15. Theconductivity (a) values changed in steps from one layer to another, as it will be shownbelow. We tested a variety of conductivity distributions (further called "a-distributions")across the layers, but within the scope of this work, we present only a few of the mostrelevant ones. As for the electrode names used here: the "hot" electrode is the mid electrodeconnected to a high-voltage (HV) source; the "grounded" electrode is electrically groundedand linked to the "hot" electrode by a switch between them; the "floating" electrode isneither connected to the HV source, nor grounded.

We used two a-distributions of the opposite nature, named for this purpose A and B. (Thenames used for these distributions are not so self-explanatory. They were internally used overthe course of this work, but we retain them in this paper for consistency reasons.) In thedistribution A, a is distributed such that the layers next to the electrodes are assigned al =amax value, while the layer in the center is assigned a8 = armin. Thus, the a-distribution issymmetrical around the central layer. In this case, am = 10-2 S/m, and 0.min = 10"10 S/m. For 1V-voltage between the electrodes, the field values for each layer of the lower line are plottedin Fig. 6 with respect to the homogeneous case (of a single, lossless slab). (The results for theupper line are omitted as they are always lower or equal to those in the lower line.) The fieldof the distribution A is lower than that of the homogeneous distribution everywhere except inthe layers next to the electrodes, where it sharply rises, which is undesirable since the fieldenhancement is the highest by the hot electrode anyway.

120 For case B (GI = Gmin, C"8 = armax), the field1151•0 A in the transient regime is moderately lower1 than in the homogeneous case everywhere100

95 except in the central layer where it has a90 Nspike. However, since breakdown can

8 2 occur in the electrostatic regime just as in34 5 6 7 9 10 1112 13 14 the transient regime, we now turn to the

gmoned dcftrc h,, situation in the electrostatic regime forFigure 6. E-field in the lower line (transient) these three distributions, which is shownin Fig. 2. We see that all three curves are bundled around the same values, except for a few

12

points near the electrodes, and in the center. That was also the case for the other distributionstested (not shown here) and we can generally state that the distribution of the field in theelectrostatic regime is quite insensitive to changes in a-distribution. Attempts to optimize thefield values based on the electrostatic regime resulted just in minor differences, which is whyin the rest of the discussion we will focus on the improvements in the transient regime.

¶70 These were possible by adjusting the form1so6 of the a-distribution function, and a max and150 a.n• values. Figure 8 shows the most1ý 140

A rain

S130 significant of these attem pts. The first120 attempt of optimization of the design,,10 - opt], is based on good properties of

1 2 3 4 5 6 T 8. . 11213.1415 models A and B. It keeps the same aqrou hot and a , but applies a form of a-function

nded electrode hot m

in which a for the inner layers (3 - 13) is

Figure 7. E-field produced by three models in based on model A, while a for the outer

the lower line in electrostatic mode. layers (1, 2, 14, 15) was based on modelB. This brought a notable improvement

over the earlier cases since the field is lower than in the homogeneous case for the mostlayers, and the field values in the outermost layers are also considerably lower comparing tothe values before.

100 Figure 8. E-field in the transient regime after9 hoig adjustment of conductivity distribution, and

.. values.

f 75 ,p[41$70

' opt3c255

501 2 3 4 5 6 7 8 9 10 11 12 13 14 15

The overall fundamental limitations in using a folded Blumlein transmission line have beenidentified and summarized.5

M. Joler, C.G. Christodoulou, and E. Schamiloglu, "Limitations to Compacting a Parallel-Plate BlumleinPulse-Forming Line," accepted and to appear in International Journal of RF and Microwave Computer-AidedEngineering (2007).

13

Liquid Breakdown Studies in Support of Task 1This research was led by Old Dominion University and is included in their portion of thisreport.

High Dielectric Constant Ceramics for Use in Blumlein Transmission Lines

Although the folded Blumlein pulser configuration described thus far is compact, it is widelyrecognized that high energy density ceramic dielectrics can further decrease the volume ofsuch pulser systems and this is what motivates our interest in breakdown in solids. Theparticular aspect of breakdown in solids that is of interest is a study of breakdown in highdielectric constant ceramics with spatially-varying permittivity (leading to a gradedresisitivity). High dielectric constant ceramics are being developed for a wide variety ofapplications in electrical engineering, including as a dielectric for high energy densitycapacitors.

The energy density y (in J/m 3) of a ceramic under the influence of an applied electric field isgiven by

y = 6o Fr E2/2 (2)

where c0 is the permittivity of free space, cr is the dielectric constant of the ceramic, and E isthe applied electric field strength. Typical candidate materials that are being studied in thisregard are TiO2 and BaTiO3.

The development of high dielectric constant ceramics is led by the ceramists, who areconcerned with the microstructure and processing techniques used in fabricating samples.Unfortunately, earlier attempts at utilizing nanocrystalline ceramics proved challenging,since the ceramists' ability to reproducible sinter large quantities of reasonably-sized sampleswas limited.

We therefore shifted our attention to working with epoxy compounds that were impregnatedwith nanocrystalline powders. This was a collaborative effort with TPL Inc. (Albuquerque,NM). This ceramic/epoxy composite material which has a high dielectric constant isproprietary in which the nature cannot be disclosed (mystery material).

Precision ball bearings were used as the electrodes to help minimize electrical fieldenhancement that would create surface flashover. One part consists of 16 samples with theelectrodes attached to the mystery materials. When testing, the material would be submersedin high voltage dielectric oil to prevent arcing between the bearings since the bearing areclose together and voltages would reach in the realm of 80 kV. There is a gap between theelectrodes where the thickness is calculated through measurements. When the ceramic/epoxycomposite material was being made, the ball bearings were temporarily glued to a fixture tocreate the gap between the precision ball bearings. Pictures of the mystery material can beviewed below (part - BB#8) in Figure 8.

14

In characterizing the electrical breakdownstrength of the mystery material, there

*1 A I * 0 were many tests that could have been0 A performed. Due to the limited number of

(d) ,T:op;, (b) .-.=ne (Cl- I ,l.e samples, the types of tests to be performedFigure 8. Photograph of precision ball depend on the desired future application ofbearings impregnated in the epoxy matrix, the material. Parts - BB#3, BB#4, and BB

#6 underwent a ramp test. Test data can beviewed below. The ramp test was at 20kV/s until the material broke down.

Tables 1-3 summarize the test data for three samples. With the data combined consideringthe different thickness between the precision ball bearings (gap containing the mysterymaterial), it was calculated that the average electrical field breakdown was 6.47 kV/mil witha deviation of 0.36 kV/mil or 2.55 MV/cm with a deviation of 0.14 MV/cm.

Sample # kV/mil MV/cm Thickness V (W) Table 1. Sample data for BB#3. Using Weibull

1 591 2.33 67.99 statistics, it was calculated that the average2 6.02 11.5 69.26 breakdown voltage was approximately 74.24 kV3 6.4033 2A9 12 _75.94 and the standard deviation was approximately

4 2.52 12 0.02 14.27 kV.5 6.42 2.53 13 83.48

.. ...7 ... .... ...... .6: 1 ....... ...... 2 M6 .....i. . 1 2 7. 3 4 ....8E 6,76 ý264 13 87.109 6,75 2.66 i 10 6 67.49

10 6.89 2.71 1 12 82.67

Sample # kV/mil I MV/cm Thickness ! V(kV) Table 2. Sample data for part - BB#6. Using(mils) S 897Weibull statistics, it was calculated that the

2 .. 6.20 2.44 . 12.5 i 77.52 average breakdown voltage was6.23 2.45 12 , 74.73 approximately 76.10 kV and the standard

4 6.45 [ 2-54 12 1 77.42S....... • ..... . ..... _ _ -• ....... . ...... ..... ....... .. ... • ..iii•.ii.•`iiii~ 657 •ii~ii?.25~`[ii~ii9•iiii 1i-]2 2.11I- 7888lli• i2 deviation was approximately 14.60 kV.

6 6.76 2.68 13 87.90

1 7 7.29 1 2.87 12 87.42

Table 3. Sample data for part - BB#4. UsingSample # kV/mii MV/cm I Thickness V (WV)(Mies) .... Weibull statistics, it was calculated that the

1 6.07 2,39 5.5 33-39 average breakdown voltage was2 6.09 2.40 55 3approximately 38.00 kV and the standard3 . 6.17 . .2.43 6 37.01S................ . 43..... . . ....................... ....... deviation w as approxim ately 7.30 MV4 6.43 2,53 5.5 35.36

5 6.84 2.69 6 41.04

6 7.08 2.79 6.5 46.04

The key result from testing this mystery material is that, just with other dielectrics tested(such as Mylar), the pulsed power breakdown electric field strength is lower than the quasi-DC breakdown electric field strength. This surprising result has yet to be fully understood.Continued efforts in this direction would be useful.

15

"UNM PERSONNEL(June 2005 -November 2006)

ECE FacultyDr. Edl Schamiloglu (consortium PI), ProfessorDr. Christos Christodoulou (UNM Co-PI), Professor and Chair

ECE Research FacultyDr. John Gaudet (0.5 FTE)Dr. Mikhail Fuks (0.5 FTE)Dr. Jerald Buchenauer (0.5 FTE)

ECE StaffMr. Ralph L. Terry, Laboratory Supervisor (0.5 FTE)

Graduate StudentsMiroslav Joler - Ph.D. - Design of High Power Blumlein Transmission Lines (graduated)Zhaoxian Zhou - Ph.D. - Multiresolution Basis Functions for Transient Field Calculations

Using the Time Domain Integral Equation Method (graduated)Jinhui Chen - Ph.D. - Electromagnetic modeling of transient breakdown phenomena in

liquids and gases (graduated)Kelly Hahn - M.S. - HPM generation using Disk Cathode (graduated)Paul Castro - M.S. - Ti0 2-impregnated epoxy for use in compact transmission lines (will

graduate August 2007)Marvin Roybal - M.S. - Low impedance Marx for relativistic magnetrons (will graduate

August 2007)Andrey Andreev - Ph.D. - Relativistic magnetron/ubitron with transparent cathode (will

graduate summer 2007)Sarita Prasad - Ph.D. - Relativistic magnetron with transparent cathode - rapid rise (will

graduate Dec. 2007)

Undergraduate StudentsMichael Abney - phase locking of magnetrons (graduated).

16

UNM PUBLICATIONS(Since August 2006)

1. Books

a) J. Benford, J. Swegle and E. Schamiloglu, High Power Microwaves, 2nd Ed. (Taylorand Francis, New York, NY, February 2007).

2. Journal Papers

a) H. Bosman, M. Fuks, S. Prasad, and E. Schamiloglu, "Improvement of the OutputCharacteristics of Magnetrons Using the Transparent Cathode," IEEE Trans. Plasma Sci.34, 606-619 (2006).b) M. Fuks, N. F. Kovalev, A. Andreev, and E. Schamiloglu, "Mode Conversion in aMagnetron with Axial Extraction of Radiation," IEEE Trans. Plasma Sci. 34, 620-626(2006).c) S. Xiao, J. F. Kolb, M. A. Malik, X. Lu, M. Laroussi, R. P. Joshi, E. Schamiloglu, andK. H. Schoenbach, "Electrical Breakdown and Dielectric Recovery of PropyleneCarbonate," IEEE Trans. Plasma Sci. 34, 1653-1661 (2006).d) G. Zhao, R. P. Joshi, V. K. Lakdawala, E. Schamiloglu, and H, Hjalmarson, "TiO 2

Breakdown Under Pulsed Conditions," J. Appl. Phys. 101, 026110-1-3 (2007).e) M. Joler, C.G. Christodoulou, and E. Schamiloglu, "Limitations to Compacting aParallel-Plate Blumlein Pulse-Forming Line," accepted and to appear in InternationalJournal of RF and Microwave Computer-Aided Engineering (2007).

3. Papers in Conference Proceedings

a) E. Schamiloglu, M. Fuks, H. Bosman, S. Prasad, and A. Andreev, "High-PowerMagnetron and Ubitron Powered by a Transparent Cathode," to appear in ProceedingsBEAMS 2006, (Oxford, UK, July 2006).b) M. Fuks, A. Andreev, H. Bosman, S. Prasad, and E. Schamiloglu, "ProspectiveApplications of the Transparent Cathode in Crossed-Field Microwave Oscillators,"Papers of Joint Technical Meeting on Plasma Science and Technology and Pulsed PowerTechnology, IEE Japan (Kauai, Hawaii, 7-9 August 2006), p. 61-66.c) H. Bosman, S. Prasad, M. Fuks, and E. Schamiloglu, "Improved Performance ofMagnetrons using the Transparent Cathode," Proceedings 14th Symposium on HighCurrent Electronics, (Tomsk, Russia, 10-15 September 2006), p. 346-35 1.d) A. S. Shlapakovski, S. N. Artemenko, V. A. Avgustinovich, A. I. Mashchenko, V. M.Matvienko, V. Yu. Mityushkina, I. I. Vintizenko, W. Jiang, and E. Schamiloglu, "Statusof the Development of X-Band Antenna-Amplifier: Design, Simulations, and PrototypeExperiments," Proceedings 14th Symposium on High Current Electronics, (Tomsk,Russia, 10-15 September 2006), p. 359-362.e) A.S. Shlapakovski, W. Jiang, and E. Schamiloglu, "Numerical Simulations of an X-

17

Band Antenna Amplifier Device," Proceedings 14th Symposium on High CurrentElectronics (Tomsk, Russia, 10-15 September 2006), p. 417-420.d) E. Schamiloglu, M. Fuks, H. Bosman, S. Prasad, and A. Andreev, "High-PowerMagnetrons and Ubitrons Driven by Transparent Cathodes," Proceedings of the FirstEuro-Asian Pulsed Power Conference, (Chengdu, China, 18-22 September 2006), p. Thi-05-1-6.

4. Abstracts

a) S. Prasad, H. Bosman, M. Fuks, and E. Schamiloglu, "Limitation of Leakage Currentin the A6 Relativistic Magnetron," IEEE International Conference on Plasma Science(Traverse City, MI, June 2006).b) M. Roybal, M. Abney, S. Prasad, M. Fuks, C.J. Buchenauer, K. Prestwich, J. Gaudet,and E. Schamiloglu, "Design and Optimization of a Low-Impedance Pulsed Power MarxGenerator to Drive a Relativistic X-band Magnetron," IEEE International Conference onPlasma Science (Traverse City, MI, June 2006).c) A.D. Andreev and E. Schamiloglu, "Measurement of I-V Characteristic of Shortpulse(10-15 ns) Electron Beam," IEEE International Conference on Plasma Science (TraverseCity, MI, June 2006).d) J.T. Fleming, P. Mardahl, L. Bowers, H. Bosman, S. Prasad, M. Fuks and E.Schamiloglu, "Three Dimensional PIC Simulations of the Transparent and EggbeaterCathodes in the Michigan Relativistic Magnetron," IEEE International Conference onPlasma Science (Traverse City, MI, June 2006).e) S. Prasad, H. Bosman, M. Fuks, and E. Schamiloglu, "Efficiency Enhancement in A6Magnetron with Transparent Cathode," IEEE International Conference on PlasmaScience (Traverse City, MI, June 2006).f) A.S. Shlapakovski, S.N. Artemenko, V.M. Matvienko, I. I. Vintizenko, W. Jiang, andE. Schamiloglu, "Status of the Development of X-band Antenna-amplifier: Design,Simulations, and Prototype Experiments," IEEE International Conference on PlasmaScience (Traverse City, MI, June 2006).

18

UNM RECOGNITION

1) Edl Schamiloglu elected EMP Fellow, Summa Foundation.2) Andrey Andreev, Sigma Xi outstanding graduate student award.

UNM - NEW DISCOVERIES, INVENTIONS, PATENTS

1. Patent Disclosure: M.I. Fuks, E. Schamiloglu, H. Bosman, and S. Prasad, "AnEggbeater Transparent Cathode for Magnetrons and Ubitrons and Related Methods ofGenerating High Power Microwaves," UNM-787, provisional patent filed August 25,2006.

19

Final Report

on the project

Basic Research Leading to Compact, Portable PulsedPower

Subcontract to Old Dominion University:

ODU - LIQUID BREAKDOWN AND ENERGY STORAGE

June 1, 2001 -November 30, 2006

submitted to

E. Schamiloglu, P.I.

Department of Electrical and Computer EngineeringUniversity of New MexicoAlbuquerque, NM 87131

by

Karl H. Schoenbach

Frank Reidy Research Center for BioelectricsOld Dominion University

830 Southampton Avenue, Suite 5100, Norfolk, VA 23510Tel. 757-683-2421

FAX: 757-314-2397e-mail: Schoenbach@ece.odu.edu

ODU INTRODUCTION

The research work at Old Dominion University has focused on experimental and modelingstudies on breakdown of electrical discharges in polar liquids, and their application forenergy storage and as closing switches in compact pulsed power systems. The liquids studiedwere water and propylene carbonate. While water is widely used as an energy storagemedium in fast pulse forming networks with the advantages of low cost, high availability,and ease of handling, its freezing temperature of 0°C prohibits its use in lower temperatureenvironments. An alternative candidate is propylene carbonate (C4H60 3). It is a polar liquidand has a dielectric constant of 65. Its freezing temperature is -55°C, which is more suitablefor lower temperature applications, for example, as energy storage and switch medium inairborne systems, which operate at high altitude.

ODU - LIST OF ACHIEVMENTS

In our work, which was performed in collaboration with the research group at the Universityof New Mexico (Dr. Schamiloglu), and also with our colleagues from the sister MURIconsortium at Texas Tech University in Lubbock, TX, we:

"* Determined the dielectric strength of water and propylene carbonate depending onpulse duration, covering the range from microseconds to ten's of nanoseconds. Theresults showed that the dielectric strength in a pin-plane electrode geometry dependson polarity. Whereas for negative polarity of the pin, the results confirm Martin'sempirical equation, but for positive polarity, they are higher than predicted. Thedielectric strength of water, which increases inversely with pulse duration, reaches1.8 MV/cm and 1.5 MV/cm for negative and positive pin polarity for 50 ns pulses. It

was higher (2.2 MV/cm) for propylene carbonate under the same conditions."* Explored the breakdown mechanism in polar liquids in response to a submicrosecond

(-100-200 ns) voltage pulse. It was shown, through simulations, that breakdown isinitiated by field emission at the interface of preexisting micro-bubbles. Impactionization within the microbubble's gas then contributes to plasma development, withcathode injection having a delayed and secondary role. The models developedadequately explained experimental observations of pre-breakdown currentfluctuations, streamer propagation and branching, as well as disparities in hold-offvoltage and breakdown initiation times between the anode and the cathode polarities.It was demonstrated that polarity effects basically arise from the large mobilitydifference between electrons and ions.

"* Studied the breakdown mechanism experimentally by using ultrashort optical andelectrical diagnostics. In particular, by using interferometric methods, we determinedthe spatial and temporal development of the precursors of the conducting plasma, andto determine the absolute values of the electric field by means of the Kerr effect.

"* Studied the recovery phase after breakdown in water. Whereas the physics of thebreakdown determines the important closing switch parameters such as breakdown

voltage, current rise and plasma development, recovery is important for any systemthat operates in a repetition rate mode. Experiments using pulse-probe methods andSchlieren diagnostics allowed us to follow the development of the disturbance causedby the breakdown over a time of more than milliseconds. After breakdown, theexpanding plasma column generates shockwaves first and later, a vapor bubble whichexpands for approximately 200 uts and then decays with a time constant of 1 ms. Thebubble decay time determines the dielectric recovery of the switch, as has beenshown with pulse-probe experiments.Studied the recovery phase after breakdown in propylene carbonate. The recovery ofpropylene carbonate was found to be similar to that of water, with plasma decay,shock wave emission, and vapor bubble formation, except for the very last phase,which is determined by chemical reactions: the generation and decay ofpolypropylene polymers. This restricts the time for dielectric recovery of propylenecarbonate switches to values of more than 10 ms.

Whereas these achievements have been reported in earlier annual reports, our efforts duringthe past 15 months have focused on developing triggerable, compact pulsed power systemswhich utilize the advantages of liquids as energy storage and switch medium.

High repetition-rate, compact, nanosecond pulse generator

We have built and tested a pulse generator that provides a 10 ns duration, 2 ns risetime, up to35 kV pulse amplitude across a matched load. The current rise obtained reaches values ofalmost 2 kA/ns. In addition, the pulse generator allows us to operate at repetition ratesexceeding 100 Hz. This has been achieved by using water switches in a flow-throughconfiguration. We were able to reach a repetition rate of up to 400 Hz, limited only by thepower supply. Earlier studies using the same concept have shown that it is possible to exceed

1 kHz repetition rates.

La ' Fig. 1 Photograph and cross-Load section of the all-water

pulse generator

By using flowing,deionized water with its

high dielectric constant (81) as dielectric between the parallel lines of the Blumlein PFL, wehave the additional benefit of considerable size reduction in the PFL and thermal control ofthe dielectric. The "all water" pulse generator occupies a volume of only 25 cm by 5 cm by 7cm, not counting the charging circuit and the hydraulic system. It generates peak powers ofmore than 100 MW at average powers of half a kW when operated at 35 kV.

Fig. 2 Breakdown voltage versus repetition rate

30 (inserts: electrodes and schematics of water

25 ... flow)>-" 20 - Transverse flow allows repetitive

operation at 400 Hz at an average powerW 15 Flowrate of160 W.M 60 mI/s0 10 - At 400 Hz, the average breakdown

5 -voltage decreases by only 13% as

Water switch compared to the breakdown voltage at

1 10 100 400 10 10 Hz.

Repetition Rate (Hz) Higher repetition rate (higher averagepower) is possible with increased flow

rate.

Electrical Triggering of Water Switches

Triggering breakdown in liquid switches has been achieved with lasers and also with fielddistortion electrodes. The concepts of laser triggering are straightforward: they are basedeither on the generation of a preionized channel between the electrodes, or the vaporizationof the liquid in the point of the laser beam. For field distortion, the trigger electrode is placedclose to the ground electrode along an equal-potential surface. When the trigger electrodereaches high voltage potential, the voltage between the trigger electrode and the electrodewith the same polarity decreases, and that of the opposite electrode increases. Thebreakdown of the overstressed second gap forces the full switch potential to appear acrossthe first gap resulting in complete closing of the switch.

+Trigger, I ElectodeGlss Fig. 3. Concept of a water switch+TigrElectrode Glass (r1)trgrnSZ• trigatron

d1.Vdo, v! /t ', Vv-1.

"7 [Water (s2)-H /GND

Triple Point We have focused on a different

triggering approach that relies on electrically generated positive streamers. The trigatronwater switch, which utilizes a trigger electrode with a triple point, has been shown to triggerthe water switch with a lower trigger voltage than the field distortion trigger concept. It isdue to the initiation of the trigger streamer from a triple point, where the electric field canreach extremely high values, even at moderate voltages. From the triple point, positivestreamers send out two filaments that bridge the gaps to the high voltage electrode andground electrode, providing a conductive path between the electrodes.

250

200

(I)150

.)1000

Trigger Voltage: 14.5 kV

16 18 20 22 24 26 28 30 32

Main Gap Voltage (kV)-Fig. 4a (left) Streamer development in the water switchtrigatron (the first photograph in this series shows the

Exposure Time 0.5 ns electrode geometry). 4b (above) Delay (and jitter) oftrigatron switch versus switch voltage.

A manuscript which discusses the concept and the results of our water switch trigatron researchis under preparation.

ODU PUBLICATIONS(Since August 2006)

Publications in Refereed Journals

a) Y. Sun, S. Xiao, J. A. White, J. F. Kolb, and K H. Schoenbach, "Compact, Nanosecond,High Repetition-Rate, Pulse Generator for Bioelectric Studies," submitted to IEEE Trans.Dielctrics and Electrical Insulation

b) J. Qian, R. P. Joshi, K. H. Schoenbach, J. R. Woodworth and G. Sarkisov, "ModelAnalysis of Self- and Laser-Triggered Electrical Breakdown of Liquid Water for PulsedPower Applications," IEEE Trans. Plasma Science 34, 1680-1691 (2006).

c) S. Xiao, J. Kolb, M. A. Malik, X. Lu, M. Laroussi, R. P. Joshi, K. H. Schoenbach,"Electrical Breakdown and Dielectric Recovery of Polar Liquids," IEEE Trans. PlasmaScience 34, 1653-1661 (2006).

d) J. Qian, R. P. Joshi, E. Schamiloglu, J. Gaudet, J. R. Woodworth, and J. Lehr, "Analysis ofthe Polarity Effects in the Electrical Breakdown of Liquids," J. Phys. D 39, 359 (2006).

e) Juergen F. Kolb, Susumu Kono, and Karl H. Schoenbach, "Nanosecond Pulsed ElectricField Generators for the Study of Subcellular Effects," Bioelectromagnetics J. 27, 172-187(2006).

Publications in Conference Proceedings

a) E. Schamiloglu, K. H. Schoenbach, R. Vidmar, R. P. Joshi, M. Laroussi, J. F. Kolb, and S.Tyo, "Compact, Portable Pulsed Power - Lessons Learned and Quo Vadis," to appear in theConf. Record of the 2006 IEEE Power Modulator Conference, Washington DC, May 14-18,2006.

b) Tammo Heeren, Juergen F. Kolb, Shu Xiao, Karl H. Schoenbach, and Hidenori Akiyama,"Pulsed Power Generators and Delivery Devices for Bioelectrical Applications," to appear inthe Conf. Record of the 2006 IEEE Power Modulator Conference, Washington, D.C. May14-18, 2006.

c) Y. Sun, S. Xiao, J. F. Kolb, K. H. Schoenbach, "A Nanosecond, High Repetition-Rate, AllWater Pulse Generator," to appear in the Conf. Record of the 2006 IEEE Power ModulatorConference, Washington, D.C. May 14-18, 2006.

d) S. Xiao, Y. Sun, J. Kolb, U. Pliquett, T. Heeren, H. Akiyama, and K. H. Schoenbach,"Electrically Triggered Water Switches," to appear in the Conf. Record of the 2006 IEEEPower Modulator Conference, Washington, D.C. May 14-18, 2006.

Final Report 21 Dec 2006

IMPROVEMENTS TO COMPONENTS FOR USE IN COMPACT PULSEDPOWER SYSTEMS

By: ROBERT J VIDMAR

Prepared for:

Professor Edl Schamiloglu, MURI Principal InvestigatorUniversity of New MexicoECE Department - Pulsed Power, Beams and Microwaves LaboratoryAlbuquerque, NM 8713 1-0001

CONTRACT NUMBER: UNM Subaward 3-19531-7820Based on AFOSR F49620-01-1-0354

Approved for Public Release: Unclassified with Distribution Unlimited

University of Nevada, Reno1664 North Virginia StreetReno, Nevada 89557-0240 USA

UNR ABSTRACT

Research on the overall effort has been on exponential transmission lines with high-impedance ratios and high-pressure liquid cooling of active and passive electroniccomponents. The most recent work addresses transmission-line design and practical RF linetermination to permit pulsed operation with nanosecond or better rise time. A geometricalalgorithm is used to provide a line shape that incorporates an exponential variation in lineimpedance and maintains a constant phase front throughout the transmission line. Thismodification permits line operation with impedance ratios exceeding 10 with a testtransmission line operating with an impedance ratio of 100. Line termination with a RF loaddistributed along the output interface is critical to observing ideal voltage step-downperformance. Driving the low-impedance end of the line requires a uniform distribution offeed points along the low-impedance end of the line. Practical realization of the necessaryhardware is difficult.

27

UNR INTRODUCTION

Transmission-line theory has been developed and applied to many, applications. A subset oftransmission line theory that deals with lines with nonuniform properties has receivedsignificant attention but some devices such as an exponential transmission line remaintheoretical' without discussion on practical limits to line performance. A basic property ofan exponential line is its performance as an ideal high-pass transformer. An earlyinvestigation by Johnson2 provides clear insight on exponential line operation such asvoltage step-up and impedance transformation but provides no insight on practical limits ofperformance. Latter investigators provide analytic expressions of circuit parametersincluding two-port S parameters3-5. Agreement between theory and experiment is very good

6for short transmission lines with impedance ratios of less than thirty . Such devices arerelatively compact and have been used extensively as impedance matching devices forcouplers. Although the S-parameters provide a complete theoretical analysis for smallimpedance ratios, they provide no answers to the practical limits experienced whenconstructing an exponential line with impedance transformations ratios greater than thirty.

Basic theory for an exponential line has a sound theoretical footing and Vidmar and Faretto 7

provide an explanation and experimental findings that suggest the limitations to exponentiallines with high impedance ratios can be overcome with a modified exponential shape. Amajor constraint on the exponential-line shape is that all rays in a propagating wavefrontpropagate at the same speed and the phase is uniform along a curved wavefront. Traditionaltheory maintains a planar wavefront resulting in significant interference for lines withimpedance ratios greater than thirty. Zucker et a18 demonstrated a long exponential line withthe thickness of the line above the ground plane tapered. That taper over part of the line inconjunction with a gentle increase of line width of a different part of the line resulted in anoverall impedance ratio of 200. The Zucker line could be made more compact byincorporating the modified exponential shape and applying it along the entire transmission-line length.

Exponential lines provide a means of transforming a low-voltage high-current pulse into ahigh-impedance high-voltage pulse. Several lines can be combined in a parallel-seriesconfiguration using two-port network theory. One implementation is to combine severallines with the low-impedance sides driven in parallel and the high-impedance ends arrangedin series. This topology provides a method of combining high-voltage pulses to obtain a veryhigh-voltage pulse. Research on a exponential line with an impedance ratio of 100 hasprovided a means of achieving compact lines with high impedance ratios.

28

UNR - LIST OF ACHIEVEMENTS

A. Exponential Transmission Line Modeling

The exponential line model 7 was used to design an exponential line with an impedance ratioof 100. A ray tracing algorithm was developed to maintain the phase of a propagatingwavefront and to adjust the arc length of the wavefront to increase in an exponential manner.The resulting algorithm was programmed to generate a taper that approximates anexponential line. The exponential line has a high-impedance of 50 Q and a low-impedanceof 500 mg). The line was fabricated from a thick copper sheet ground plane and 1-mm thickTeflon as a dielectric layer between the upper transmission line surface and the ground plane.The geometry was fine tuned for minimum backscatter from the high-impedance (50 QŽ) endusing the SII parameter of a vector network analyzer. Measurements were conducted withcontinuous signals at discrete frequencies above the high-pass cutoff frequency and for a 2-ns duration pulse. These experiments yielded an approximate voltage step-down ratio of 10.

B. RF Power Termination

The low-impedance end of the line was terminated with multiple resistors closely spaced.The number of resistors and their value were computed based on multiple resistors inparallel. Industrial composition resistors were initially used to terminate the line. Theyprovided some results but the resistors failed to function properly at high frequency due toexcessive reactance. RF resistors were used but they were not rated for GHz operation. Theresults presented by Vidmar and Faretto7 used these resistors and strongly suggested that thetermination resistors need to have a multi-GHz rating for proper termination. RF resistorswere investigated and vendors identified. A quantity of 40-2 RF resistors were customordered from Mini-Systems Inc. Their series WAMT 1AT-40ROOG-T resistors are rated for20 GHz operation with 2% tolerance and are tinned for solder application. These resistorshave a length of 40 mils or approximately 1 mm.

"A technique was developed to solder these resistors to a small copper ground plane test plate."A pre-warming fixture in conjunction with a 300 W soldering iron was necessary to meltsolder. Soldering resistors involved tinning the test plate, applying a rosin flux, hot-airdrying the rosin till it was tacky, picking up the resistor with a vacuum pick-up tool, andpositioning it on the tacky rosin. With the resistor in place and the plate pre-warmed, it waspossible to apply heat from the 300-W soldering iron approximately 1 to 2 inches from theresistor. After a few seconds of heating, a melting wave front appears and propagates to theresistor. When the wavefront hits the resistor, capillary attraction sucks solder to the resistor,fixing it in place. Application of this technique on the massive transmission line failed,because differential thermal expansion of the copper plate resulted in significant buckling. Alarge Teflon sheet that covers the copper ground plane and serves as the dielectric layer inthe transmission line is also heated resulting in additional buckling at the Teflon-to-copperinterface. The small resistors were attracted to the small gaps between the copper ground

29

plane and the Teflon sheet via capillary attraction of the liquid rosin and solder. Theresistors moved out of position with some of them appearing to get sucked under the Teflon.

The soldering approach is very difficult because of buckling of the copper sheet and theTeflon. Fixturing individual resistors in place and soldering were considered, but that wouldnot alleviate buckling and the prospect of permanent mechanical distortion of the groundplane and Teflon sheet. Uniform heating of the entire end of the transmission line in a largeoven would alleviate buckling but the infrastructure to do so was not available. An approachusing conductive epoxy alleviates the buckling and distortion issue noted but requires aspecial surface treatment of the ends of the resistors for proper RF operation. Lead time toprocure appropriate resistors was approximately three months and the research effortterminated before a solution to this problem could be implemented.

30

UNR - SUMMARY OF RESULTS

The major research results are the following:

"* Quantify Theoretical Limits of Exponential Lines with High Impedance Ratios.

"* Quantify Performance of High-Impedance Transmission Line.

"* Utilization of RF Chip Termination Resistors.

A. Limitations in Implementation of Exponential Line Theory

The ray propagation model provides a refinement in the mechanical design of an exponentialline with a high-impedance ratio. This refinement provides a method of designing a compacttransmission line by combining a dielectric taper as well as expanding the line width at thelow-impedance end of the line to achieve maximum impedance ratio with minimum length.The limitation on minimum line length and maximum line impedance ratio is driven by phasevariations at the low-impedance end of the line and material properties. At the high-impedance end of the line dielectric breakdown voltage and width of the current conductorprovides a practical limit of approximately one thousand ohms. At the low-impedance end ofthe line, the line is wide and the dielectric can be thinner because the voltage is lower. Thethickness of dielectric can be tapered down by a factor of the square root of the impedanceratio. A typical value of fifty milliohms may be achieved at the low-impedance end of theline. A total impedance ratio of one thousand should be possible with a very hard to achieveupper limit of approximately ten thousand. The experiments conducted provide a practicalinput on the overall feasibility of constructing compact exponential lines with high-impedance ratios.

B. Results from an Exponential Line with a 100-to-1 Impedance Ratio

The investigation of a 100-to-i impedance ratio exponential line demonstrates the feasibilityof an approach to maintain coherent phase fronts at the wide low-impedance end of anexponential line. Tuning the line in real time for minimum reflections using a networkanalyzer set for SlI was very effective. When SlI is displayed in time-domain mode, smallirregularities along the length of the line were easily identified and corrected. The resultsdemonstrated an approximate 10-to-i voltage step-down ratio in correspondence to theory.Pure resistive termination at the low-impendence end is necessary for proper high-frequencyoperation and minimization of internal reflections within the line. When driving the linefrom the low-impedance end, it is important to have nearly ideal components that have aminimum of reactance at the frequencies of interest.

31

C. Utilization of RF Chip Termination Resistors

The overall experience in working with 1-mm long RF termination resistors is that solderingis a difficult activity and is not well suited to large thick surfaces. If soldering is to beutilized in fixing resistors, a thin copper sheet for a ground plane is preferred to a thick sheet.An alternative approach is to use RF chip resistors specified with an end treatment which iscompatible with conductive epoxy adhesives. The latter approach eliminates the thermalexpansion issue of buckling associated with direct soldering and provides additional finessein final placement of resistors.

32

[ UNR PERSONNEL

Personnel. The primary personnel on this project has been

e Robert J Vidmar, Principal Investigator

Additional personnel that contributed to the project

"* Harold Faretto, Electronics Technician

"* Andrew Oxner, Development Technician

"* Quinn Sinnott, Undergraduate Student Worker

Interactions. The principal research has been presented at the 15 th International PulsedPower Conference, Monterey, CA 2005. There were several interactions at theseconferences with representatives from DoD Laboratories, National Laboratories, privatecompanies, and Universities from around the world.

UNR PUBLICATIONS

Vidmar, R, and H Faretto, "Exponential Transmission Line Design for Impedance RatiosGreater than 10, 15th International Pulsed Power Conference, 13-17 June, paper10120, Monterey, CA, 13-17 June 2005.

33

UNR REFERENCES

[1] G Miano and A Maffucci, Transmission Lines and Lumped Circuits, Academic Press,San Diego, CA, pp 271-279, 2001.

[2] W. C. Johnson, Transmission Lines and Networks, McGraw-Hill, New York, pp. 201-208, 1950.

[3] V. Ramachandran, and K. K. Nair, "Equivalent Circuits of an ExponentialTransmission Line", IRE Trans. on Circuit Theory, vol. CT-7, pp. 71- 74, 1960.

[4] S. C. Dutta Roy, "Low-Frequency Wide-Band Impedance Matching by ExponentialTransmission Lines," Proc. of the IEE, vol. 67, pp. 1162-1163, 1979.

[5] R. J. Vidmar, 14th International Conference on High-Power Particle Beams, BEAMS2002, June 23-28, 2002, AlP Conference Proceedings, vol. 650, pp. 41-44, 2002.

[6] R. P. Arnold and W. L. Bailey, Electronic Design, vol. 22, 136-139 (1974).

[7] R. Vidmar and H. Faretto, "Exponential Transmission Line Design for ImpedanceRatios Greater than 10", 151h International Pulsed Power Conference, 13-17June, paper 10120, Monterey, CA, 13-17 June 2005.

[8] 0. S. F. Zucker, D. Giorgi, K. Stuurman, K. Cardwell, P Solone, and M. Bonebright,"Design and Experimental Performance of a Picosecond Rise-Time AdiabaticTransformer with Impedance Ratio of 200:1," Pulsed Power Conference 1993,Digest of Technical Papers, pp 297-299, 9 th IEEE International, Vol 1, 21-23June, 1993.

34